EP2766739B1 - Procedure and device for detecting insulation faults in a low voltage network - Google Patents

Description

The present invention relates to a device and a method for detecting, in an electrical network carrying a low-voltage alternating voltage, an insulation fault resulting in the appearance of partial discharge signals. The present invention relates in particular to a device and a method for detecting a fire initiation.

It is known that a fire initiation in a power line results in the prior appearance of partial discharges which are the sign of a lack of insulation of the power line. These partial discharges lead to the ignition of surrounding materials without conventional safety devices being able to operate. A differential circuit breaker is indeed inoperative at this stage of the fault because no leakage to the ground is still detectable. Similarly, a power circuit breaker is not affected because the powers involved remain modest although the ignition energies are high.

Such partial discharges create pulsed electromagnetic disturbances, termed partial discharge signals, that can be detected. The detection of partial discharge signals allows the implementation of countermeasures to prevent the ignition of surrounding materials.

Methods or devices for detecting partial discharge signals have already been proposed. The US Patent 6,172,862 describes a detection system comprising a sensor capable of measuring discharges of the order of a few picocoulombs and means for comparing them with a reference threshold. The patent application US 2010/0073008 discloses a detection system comprising a filter configured to transmit high frequency partial discharge signals to a differential amplifier and to filter the low frequency voltage of the network. The patent application US 2008/0309351 discloses a detection device comprising a first branch having a first impedance for partial discharge signals occurring in the line voltage, and a second branch having a second impedance lower than the first for the partial discharge signals occurring in the line voltage; so that the partial discharge signals are preferentially transmitted in the second branch. The US Patent 5,835,321 discloses a detection device comprising a band pass filter and configured to detect a partial discharge current in a determined frequency band.

Otherwise, EP 0 639 879 A2 teaches electric arc detection by detecting changes in the cycle of an alternating current. The alternating current is sampled with a notch filter during each of a plurality of cycles of the alternating current, to provide a plurality of currents In, m, where "n" is the phase and "m" is the cycle. Absolute values of current difference signals are accumulated in a synchronous summer during the m cycles. A signal indicative of the presence of an arc is generated in response to given conditions and taking into account combinations of conditions of the absolute cumulative current of the difference signals.

JP 2005 156452 refers to gas-insulated switchgear (GIS) devices using SF6-type gas as a means of isolating two high-voltage elements, and is intended to discriminate several types of partial discharges , internal and external, by means of a sensor or an antenna.

There are therefore various known methods for the detection of partial discharge signals, with or without filtering, by current or voltage detection, variable threshold, etc. These various methods generally aim at distinguishing "false" partial discharge signals, which are not related to an insulation fault, from "true" partial discharge signals, in order to avoid false alarms. Indeed, various spurious signals that may appear on the network are not due to an insulation fault resulting in fire initiation or other malfunction to be detected, and are only caused by devices or lighting connected to the network, such as lighting neon. Methods based on frequency analysis of partial discharge signals and / or fine determination of detection thresholds are sometimes ineffective in differentiating true and false partial discharge signals.

It may therefore be desirable to provide a method and a device for detecting an insulation fault which offers a better detection reliability.

Embodiments of the invention relate to a detection device, in an electrical network carrying a low-voltage alternating voltage (Vac), of an insulation fault between a phase wire (P) and a neutral wire (N). ) resulting in the appearance of partial discharge signals (PDS), comprising means for detecting partial discharge signals by voltage analysis (Vac) or the current carried by the electrical network, and means for associating with each partial discharge signal detected a parameter indicative of the position of the partial discharge signal on the AC voltage curve, and to determine if there exists, between several partial discharge signals detected during cycles of variation different from the AC voltage, to minus an indicative correlation of an insulation fault, taking into account the parameter indicative of the position of the partial discharge signal on the alter voltage curve native.

According to one embodiment, the device is configured to define as a parameter indicative of the position of a partial discharge signal on the curve of the AC voltage, groups of voltage values of the AC voltage, and to associate with each signal of discharge detected a group of voltage values of the AC voltage.

According to one embodiment, the device is configured to identify each group of voltage values of the AC voltage by means of a time slot representing a fraction of the period of the AC voltage.

According to one embodiment, the device is configured to detect an insulation fault when a first number of partial discharge signals is detected during a first number of successive cycles of the AC voltage in relation to the same group of voltage values. of the alternating voltage.

According to one embodiment, the device is configured to detect an insulation fault when a first number of partial discharge signals is detected during a first number of successive cycles of the AC voltage in relation to a first group of voltage values of the alternating voltage, and a second number of partial discharge signals is then detected during a second number of successive cycles of the AC voltage in relation to a second group of voltage values of the AC voltage comprising voltage values lower than the values of the AC voltage. first group.

According to one embodiment, the partial discharge signal detection means comprise an acquisition circuit for extracting from the AC voltage a reference signal capable of containing partial discharge signals, a reference signal processing circuit, configured for, for each group of voltage values of the AC voltage: determining a current value of the amplitude of the reference signal, calculating an average value of the amplitude of the envelope reference signal from the current value of the the amplitude of the reference signal and M-1 preceding values of the amplitude of the reference signal measured during previous cycles of the AC voltage in relation to the group of voltage values of the AC voltage considered, to determine a threshold of detecting which is a function of the average value of the amplitude of the reference signal, and detecting a partial discharge signal if the current value of the amplitude of the reference signal is greater than the detection threshold.

According to one embodiment, the acquisition circuit comprises an integrator circuit receiving as input a filtered signal extracted from the AC voltage and supplying the reference signal.

According to one embodiment, the detection circuit is configured to digitize the reference signal for each group of voltage values of the AC voltage, and to provide a digital sample representative of the amplitude of the reference signal in relation to the group of voltage values of the considered AC voltage.

According to one embodiment, the detection circuit is configured to calculate the average value of the amplitude of the reference signal from a group of M digital samples comprising the digital sample representative of the current amplitude of the reference signal. and M-1 digital samples representative of the amplitude of the reference signal during previous cycles of the AC voltage in relation to the same group of voltage values of the AC voltage.

According to one embodiment, the detection circuit is configured to generate N-bit words in which each bit is associated with a group of voltage values of the AC voltage, and has a first value if a partial discharge signal has has been detected in relation to the group of voltage values of the AC voltage considered and a second logic value if a partial discharge signal has not been detected in relation to the group of voltage values considered.

Embodiments of the invention also relate to a method for detecting an insulation fault between a phase wire (P) and a neutral wire (N) in an electrical network carrying a low voltage alternating voltage (Vac). , comprising steps of detecting partial discharge signals by analysis of the voltage (Vac) or of the current carried by the electrical network, and steps of associating with each detected partial discharge signal a parameter indicative of the position of the signal of partial discharge on the curve of the AC voltage, and to determine if there exists, between several partial discharge signals detected during cycles of variation different from the AC voltage, at least one correlation indicative of an insulation fault, taking into account the parameter indicative of the position of the partial discharge signal on the curve of the AC voltage.

According to one embodiment, the method comprises the steps of defining as a parameter indicative of the position of a partial discharge signal on the AC voltage curve, groups of AC voltage voltage values, and associate with each detected discharge signal a group of voltage values of the AC voltage.

According to one embodiment, each group of voltage values of the AC voltage is identified by means of a time slot representing a fraction of the period of the AC voltage.

According to one embodiment, an isolation fault is detected when a first number of partial discharge signals is detected during a first number of successive cycles of the AC voltage in relation to the same group of voltage values of the AC voltage. .

According to one embodiment, an insulation fault is detected when a first number of first partial discharge signals is detected during a first number of successive cycles of the AC voltage in relation to a first group of voltage voltage values. alternative, and a second number of second partial discharge signals is then detected during a second number of successive cycles of the AC voltage in relation to a second group of voltage values of the AC voltage comprising voltage values lower than the values of the first one. group.

• la figure 1 représente un exemple de réalisation d'un dispositif selon l'invention de détection d'un défaut d'isolement dans un réseau électrique,
• les figures 2A à 2F représentent des signaux apparaissant dans le dispositif de la figure 1, et
• la figure 3 représente des étapes d'un procédé de détection d'un défaut d'isolement mis en oeuvre au moyen du dispositif de la figure 1.

Embodiments of the invention will be described in the following nonlimitingly, in connection with the attached figures in which:

• the figure 1 represents an exemplary embodiment of a device according to the invention for detecting an insulation fault in an electrical network,
• the Figures 2A to 2F represent signals appearing in the device of the figure 1 , and
• the figure 3 represents steps of a method for detecting an insulation fault implemented by means of the device of the figure 1 .

The figure 1 represents a DTC device according to the invention for detecting an insulation fault in an electrical network. The DTC device is connected here to an electrical network comprising an N-neutral wire and a P-phase wire, and is configured here to detect fire initiation. The network carries an alternating voltage Vac, for example a voltage of 240 V at a frequency of 50 Hz.

The DTC device comprises an acquisition circuit ACT, a processing circuit MC, a ZCTT circuit for zero crossing detection ("Zero Crossing Circuit"), and a PSCT power supply circuit.

The circuit ZCCT receives the voltage Vac and supplies a logic signal ZC having a determined variation front at each zero crossing of the voltage Vac, for example a rising edge (transition from 0 to 1). The PSCT supply circuit also receives the voltage Vac and supplies a DC voltage Vdd to the processing circuit MC. The acquisition circuit ACT extracts from the alternating voltage Vac a signal Ve3 capable of containing partial discharge signals. The processing circuit MC analyzes the signal Ve3 to detect a fire initiation.

The acquisition circuit ACT comprises a coupler circuit CCT receiving the voltage Vac and supplying a signal Ve1, a bandpass circuit BPCT receiving the signal Ve1 and providing a signal Ve2, and an envelope extracting circuit EECT receiving the signal Ve2 and providing the signal Ve3. The signals Ve1, Ve2, Ve3 are voltage signals here.

The coupler circuit CCT here comprises an isolation transformer comprising a primary coil L1 connected to the network and a secondary coil L2 providing the signal Ve1. A capacitor C1 arranged in series with the primary coil L1 and a capacitor C2 arranged in series with the secondary coil L2 give the circuit CCT a high-pass filtering function to eliminate the 50 Hz component of the voltage Vac.

The partial fire-generating discharge signals have a broad spectrum of frequencies and are therefore likely to be detected in all frequency bands. However, some frequency bands are "polluted" by signals that may appear as partial discharge signals, including signals generated by the carrier current data transmission devices. In one embodiment of the invention, it is therefore preferred to reduce the signal analysis band Ve1 by means of the band-pass circuit BPCT, so as to eliminate carrier data transmission frequencies. Another parameter dictating the choice of the bandwidth of the circuit BPCT is the duration of time slots described below. For example, the bandwidth of the BPCT circuit is 145 kHz to 1.7 MHz in Europe and 500 kHz to 1.7 MHz in the United States.

The EECT envelope extraction circuit is realized here in the form of a half-wave integrator circuit. It comprises a half-wave rectification diode D1 whose anode receives the signal Ve2. The cathode of the diode D1 is connected to ground via a capacitor C3 in parallel with a resistor R1, and supplies the signal Ve3 to the processing circuit MC. The circuit R1C3 smooths the rectified signal Ve3, which corresponds to an integration of this signal. The circuit R1C3 has a time constant RC which is less than the duration of time slots described below, for example equal to half this duration, ie 0.5 ms in relation to the example described below. The signal Ve3 is a noisy signal capable of containing partial discharge signals.

The Figures 2A, 2B, 2C and 2D respectively represent the voltage Vac, the signal Ve1, the signal Ve2 and the signal Ve3. These signals are shown schematically for illustration only and their shape is arbitrary. The signal Ve3 has a PDS peak forming a partial discharge signal.

The processing circuit MC analyzes the signal Ve3 to detect PDS partial discharge signals. It then performs a kind of "sorting" to differentiate the PDS signals that are representative of a fire initiation and those that are not. This sorting is performed by applying one or more correlation rules that take into account the position of the PDS signals on the curve of the AC voltage Vac.

For this purpose, the circuit MC defines time slots SLi ("time slots"), designated in the following "slots" ("slots"), in which it analyzes the signal Ve3. These slots are synchronized to the alternating voltage Vac by means of the logic signal ZC, and each represents a fraction of a cycle of the voltage Vac. They are here of constant duration and equal to 1 / N times the period T of the voltage Vac, N being an integer representing the number of slots per period T. Each slot SLi has a rank i determined relative to the zero crossing of the voltage Vac and thus corresponds to a group of voltage values of the alternating voltage Vac.

The signal Ve3 is here rectified semi-alternately by the envelope extraction circuit EECT, the processing circuit MC does not distinguish between the positive and negative half-waves of the voltage Vac. The duration of the slots is here equal to 1 / N times the half-period T, ie 1 / 2N times the period T.

In the exemplary embodiment shown on the Figures 2A to 2F , the parameter N is chosen equal to 10. Thus ten slots SL0 to SL9 are distinguished by half-period of the voltage Vac, the first slot SL0 being placed after the zero crossing of the voltage Vac, the last slot SL9 being placed before the next zero crossing of the voltage Vac. Each slot SLi has a duration of 1 ms if the voltage Vac has a frequency of 50 Hz and a period T of 20 ms.

The processing circuit MC is here a microcontroller equipped with a sampling and digitizing input of the signal Ve3. The signal processing Ve3 is therefore performed on digital samples Vi of this signal, represented on the figure 2E . The sampling is carried out for example at a frequency of 100 kHz, ie 100 samples Vi, j of the signal per slot, where i is the rank of the slot considered and j the rank of the sample in the slot in question. The microcontroller returns these 100 samples Vi, j to a single sample Vi per slot, for example by choosing as sample Vi the sample Vi, j having the highest value of the 100 samples Vi, j, or by calculating the average value 100 samples Vi, j.

• une phase de détection de signaux PDS, et
• une phase de détection d'un amorçage d'incendie, ou phase de corrélation, visant à mettre en évidence une corrélation représentative d'un amorçage d'incendie entre les différents signaux PDS détectés.

The digitized and sampled Ve3 signal processing comprises two main phases:

• a PDS signal detection phase, and
• a fire ignition detection phase, or correlation phase, for highlighting a correlation representative of a fire initiation between the various detected PDS signals.

Here, the term "PDS signal" or "PDS partial discharge signal" denotes any signal having the profile of a "true" partial discharge signal but which may have an origin other than an insulation fault which may lead to an ignition failure. 'fire. It may be for example signals emitted by lighting systems or equipment connected to the network. The distinction between "true" and "false" partial discharge signals is made during the correlation phase.

Partial discharge signal detection phase

The PDS signal detection phase is illustrated by the flowchart on the figure 3 and comprises steps E1 to E11.

In step E1, the microcontroller is informed of the zero crossing of the alternating voltage Vac by changing the signal ZC to 1. In step E2, the microcontroller resets a count value CMP to 0. This value is constantly incremented by a counter of the microcontroller clocked by a clock signal (not represented on the figure 1 ). This count value allows the microcontroller to distinguish the different slots. The microcontroller also sets to 0 the index i of the slots after zero crossing of the voltage Vac.

Steps E3 to E8 are repeated by the microcontroller for each slot SLi, starting with slot SL0. In step E3, the microcontroller samples and digitizes the signal Ve3 to take a sample Vi representative of the amplitude of the signal Ve3 in the slot considered SLi. The sample Vi is determined as indicated above, by selecting the largest of the samples Vi, j or by calculating the average value of the samples Vi, j in the slot SLi.

In step E4, the microcontroller calculates a mean value MVi of the signal Ve3 in the slot considered SLi, from M samples Vi. These M samples Vi comprise the current sample Vi which has just been taken and M-1 samples Vi taken in this same slot during previous cycles of the AC voltage Vac. Like the previous M-1 samples, the current sample Vi is also stored in order to allow the microcontroller to recalculate the average value MVi at the next half-period of the voltage Vac. The oldest of the M-1 samples is erased so as not to saturate the memory of the microcontroller. The number M is for example equal to 1000, which involves storing 1000 samples Vi per slot or 10000 samples in total for the slots SL0 to SL9.

In step E5, the microcontroller calculates the standard deviation Ei of the M samples used for calculating the average value. In step E6, the microcontroller calculates a dynamic detection threshold Si = MVi + K * Ei by adding K times the standard deviation to the average value MVi, K being at least equal to 1. In step E7, the microcontroller determines whether the sample Vi is greater than the dynamic threshold Si. If so, the microcontroller considers that a partial discharge signal has been detected in the slot in question and sets a bit bi of rank i corresponding to the rank i of the slot considered SLi. If not, the microcontroller considers that a partial discharge signal has not been detected and sets the bit bi to 0.

The value K * Ei corresponds to an acceptable variation interval of the signal Ve3 relative to its mean value MVi in the slot considered. A sample Vi having a value below the threshold MVi + K * Ei is not considered as representative of the presence of a PDS signal. This range of variation allows the microcontroller not to be blinded by jumps in values of the samples Vi which are statistically acceptable and located on the central part of a Gaussian curve of the variations of the signal Ve3. The coefficient K may be chosen equal to 3, which means that signal variations Ve3 less than 3 times the standard deviation Ei in the slot under consideration will not be considered as PDS signals. Those skilled in the art will of course be able to retain other means of determining the acceptable range of variation, which are not based on a standard deviation calculation.

In step E8, the microcontroller consults the count value CMP and determines whether it is greater than a number D corresponding to the duration of a slot. If not, the microcontroller repeats step E8 until the number D is reached.

In step E9, the microcontroller determines whether the current slot of rank i is the last slot of the half-period of the voltage Vac, by determining if i = 9.

If not, the microcontroller goes to step E10 where it increments the rank i of the slot (i = i + 1), resets the count value CMP, and then repeats steps E3 to E9.

If so, the microcontroller goes to step E11 where it generates a binary word Wk comprising all bits bi configured in step E7 for each of the slots, here Wk = b0b1b2b3b4b5b6b7b8b9. The microcontroller then returns to step E1 to detect the zero crossing of voltage Vac and repeats steps E2 to E11, and so on.

By way of illustration, two half-periods 1 / 2T _k-1 and 1 / 2T _k are represented on the Figures 2A to 2F . The figure 2E shows the value of the samples Vi in each slot SL0 to SL9 as well as the values of the dynamic detection thresholds S0 to S9 which are calculated by the microcontroller for each slot during each step E6. It appears that during the half-period 1 / 2T _k-1, none of the calculated samples Vi has a value greater than the corresponding detection threshold Si. The corresponding binary word W _k-1 therefore comprises only bits with 0. During the next half-period 1 / 2T _k-1 , only the sample V3 of the fourth slot SL3 has a value greater than the corresponding detection threshold S3. The fourth bit b3 of the corresponding binary word W _k is therefore equal to 1 while the other bits are at 0, the word W _k thus being equal to 0001000000.

The person skilled in the art will note that the embodiment of the method of the invention which has just been described has been presented in the form of a recursive flow diagram for the sake of simplification of the description. This method may comprise in practice multitasking processing steps that are almost simultaneous. In particular, the step E1 for detecting the zero crossing of the voltage Vac, the steps E2 and E10 for resetting the count value, and the step E8 for monitoring the count value can be executed in a loop during the duration of a slot and simultaneously with the execution of calculation steps E4 to E7. Similarly, the calculation steps E4 to E7 can be performed in deferred time if the sampling step E3 extends for the duration of the slot in question, in particular when the microcontroller takes 100 samples Vi, j per slot and determines the value of the sample Vi.

Also, the fire detection detection phase which will now be described can be conducted simultaneously with the PDS signal detection phase, with an offset of half a period or more. For example, while the microcontroller conducts the phase of detecting PDS signals over half a period of the voltage Vac, it can simultaneously execute the detection phase of a fire initiation based on the detected PDS signals. during the previous half-periods.

Detection phase of a fire initiation

The search for a correlation representative of a fire initiation between the different detected PDS signals is made taking into account their position on the curve of the AC voltage Vac, here taking into account the slots SLi in which they were detected. The microcontroller has one or more correlation rules COR1, COR2, COR3, ... determined by tests and experiments. An alert is triggered when at least one of these correlation rules is satisfied. When an alert is triggered, the microcontroller implements a countermeasure to prevent a fire. For example, as shown on the figure 1 , the microcontroller can provide a warning signal SA that opens SWi switches through which the AC voltage Vac is applied to the network.

Examples of correlation rules COR1 to COR3 will be described in the following without limitation.

Example 1 - Rule COR1: An alert is triggered if a number N1 of successive PDS signals has been detected in the same slot SLi during successive cycles of the voltage Vac.

$$\mathrm{W}1=000\mathbf{1}000000$$

$$\mathrm{W}2=000\mathbf{1}000000$$

The rule COR1 is based on the assumption that the appearance of successive PDS signals in the same slot is representative of a fire initiation. If for example N1 = 3, an alert will be triggered for example when the microcontroller generates the following words (successive appearances of three PDS signals in the fourth slot (3): $$\mathrm{<figure-callout\; id="0"\; label="W"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">W</figure-callout>}0=000\mathbf{1}000000$$

$$\mathrm{<figure-callout\; id="1"\; label="W"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">W</figure-callout>}1=000\mathbf{1}000000$$

$$\mathrm{W}2=000\mathbf{1}000000$$

Example 2 - Rule COR2: a number N1 of PDS signals has been detected in the same slot SLi, during P1 successive half-periods of the voltage Vac. An alert is triggered if N1> 0.5 P1.

$$\mathrm{W}1=0000000000$$

$$\mathrm{W}2=00\mathbf{1}0000000$$

$$\mathrm{W}3=0000000000$$

$$\mathrm{W}4=00\mathbf{1}0000000$$

The COR2 rule is based on the assumption that the repetitive appearance of PDS signals in the same slot is representative of a fire initiation if these signals are close enough to each other and even if they are not successive. If for example P1 = 5, the alert threshold is N1> 2. As an example, an alert will be triggered when the microcontroller generates the following words (repetitive but not necessarily successive appearances of a PDS signal in the third slot (2): $$\mathrm{<figure-callout\; id="0"\; label="W"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">W</figure-callout>}0=00\mathbf{1}0000000$$

$$\mathrm{<figure-callout\; id="1"\; label="W"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">W</figure-callout>}1=0000000000$$

$$\mathrm{W}2=00\mathbf{1}0000000$$

$$\mathrm{<figure-callout\; id="3"\; label="W"\; filenames="imgf0002.png"\; state="\{\{state\}\}">W</figure-callout>}3=0000000000$$

$$\mathrm{<figure-callout\; id="4"\; label="W"\; filenames="imgf0002.png"\; state="\{\{state\}\}">W</figure-callout>}4=00\mathbf{1}0000000$$

Example 3 - Rule COR3: a number N1 of PDS signals has been detected in a slot SLi corresponding to a first voltage value of the AC voltage Vac, during P1 successive half-periods, then a number N2 of PDS signals has been detected in another slot SLi 'corresponding to a second voltage value of the AC voltage Vac lower than the first voltage value, for P2 half-periods successive following the P1 first half-periods. An alarm is triggered if N1> 0.5 P1 and N2> 0.5 P2.

The rule COR3 is based on the assumption that repetitive but not necessarily successive appearances of PDS signals in a first slot and then in a second slot corresponding to a lower value of the voltage Vac, is representative of a fire initiation. Indeed, a fire initiation results in a gradual destruction of the insulating sheaths and therefore a decrease in their dielectric constant at the point of breakdown. The amplitude of the voltage Vac at which the partial discharges occur therefore tends to decrease as the insulation deteriorates.

$$\mathrm{W}1=0000000000$$

$$\mathrm{W}2=0000\mathbf{1}00000$$

$$\mathrm{W}3=0000000000$$

$$\mathrm{W}4=0000\mathbf{1}00000$$

$$\mathrm{W}5=000\mathbf{1}000000$$

$$\mathrm{W}6=0000000000$$

$$\mathrm{W}7=000\mathbf{1}000000$$

$$\mathrm{W}8=000\mathbf{1}000000$$

If for example P1 = 5 and P2 = 4, the warning thresholds are N1> 2 and N2> 2. For example, an alert will be triggered when the microcontroller generates the following words (repetitive appearances of a signal PDS in the fifth slot SL4 then in the fourth slot (3): $$\mathrm{<figure-callout\; id="0"\; label="W"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">W</figure-callout>}0=0000\mathbf{1}00000$$

$$\mathrm{<figure-callout\; id="1"\; label="W"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">W</figure-callout>}1=0000000000$$

$$\mathrm{W}2=0000\mathbf{1}00000$$

$$\mathrm{<figure-callout\; id="3"\; label="W"\; filenames="imgf0002.png"\; state="\{\{state\}\}">W</figure-callout>}3=0000000000$$

$$\mathrm{<figure-callout\; id="4"\; label="W"\; filenames="imgf0002.png"\; state="\{\{state\}\}">W</figure-callout>}4=0000\mathbf{1}00000$$

$$\mathrm{<figure-callout\; id="5"\; label="W"\; filenames="imgf0002.png"\; state="\{\{state\}\}">W</figure-callout>}5=000\mathbf{1}000000$$

$$\mathrm{W}6=0000000000$$

$$\mathrm{W}7=000\mathbf{1}000000$$

$$\mathrm{<figure-callout\; id="8"\; label="W"\; filenames="imgf0002.png"\; state="\{\{state\}\}">W</figure-callout>}8=000\mathbf{1}000000$$

The correlation criteria involved in the COR2 and COR3 rules can be made more stringent by providing that the PDS signals are successive and not simply repetitive, which means that P1 must be equal to N1 in the rule COR2, which then becomes similar to the rule COR1, and to impose that P1 is equal to N1 and P2 equal to N2 in the rule COR2.

$$\mathrm{W}1=0000\mathbf{1}00000$$

$$\mathrm{W}2=0000\mathbf{1}00000$$

$$\mathrm{W}3=0000\mathbf{1}00000$$

$$\mathrm{W}4=0000\mathbf{1}00000$$

$$\mathrm{W}5=000\mathbf{1}000000$$

$$\mathrm{W}6=000\mathbf{1}000000$$

$$\mathrm{W}7=000\mathbf{1}000000$$

$$\mathrm{W}8=000\mathbf{1}000000$$

In this case, If for example P1 = 5 and P2 = 4, an alert will be triggered when the microcontroller generates the following words (successive appearances of 5 PDS signals in the fifth slot SL4 and 4 PDS signals in the fourth slot SL3): $$\mathrm{<figure-callout\; id="0"\; label="W"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">W</figure-callout>}0=0000\mathbf{1}00000$$

$$\mathrm{<figure-callout\; id="1"\; label="W"\; filenames="imgf0001.png,imgf0002.png"\; state="\{\{state\}\}">W</figure-callout>}1=0000\mathbf{1}00000$$

$$\mathrm{W}2=0000\mathbf{1}00000$$

$$\mathrm{<figure-callout\; id="3"\; label="W"\; filenames="imgf0002.png"\; state="\{\{state\}\}">W</figure-callout>}3=0000\mathbf{1}00000$$

$$\mathrm{<figure-callout\; id="4"\; label="W"\; filenames="imgf0002.png"\; state="\{\{state\}\}">W</figure-callout>}4=0000\mathbf{1}00000$$

$$\mathrm{<figure-callout\; id="5"\; label="W"\; filenames="imgf0002.png"\; state="\{\{state\}\}">W</figure-callout>}5=000\mathbf{1}000000$$

$$\mathrm{W}6=000\mathbf{1}000000$$

$$\mathrm{W}7=000\mathbf{1}000000$$

$$\mathrm{<figure-callout\; id="8"\; label="W"\; filenames="imgf0002.png"\; state="\{\{state\}\}">W</figure-callout>}8=000\mathbf{1}000000$$

It will be apparent to those skilled in the art that the present invention is susceptible to various variations and embodiments.

Thus, in one embodiment, the acquisition circuit ACT is made differently and is for example based on a current detection.

In another embodiment, the rank i of the slots is determined relative to another voltage value of the voltage Vac, for example its peak value.

In yet another embodiment, the search for a correlation between the PDS signals is provided by an external calculation circuit receiving the binary words Wk provided by the microcontroller.

In still another embodiment, the processing circuit MC, instead of being a microcontroller, comprises an analog processing chain configured to detect the PDS signals and provide, whenever a PDS signal is detected, a signal analog whose value or amplitude is representative of the time elapsed between the time when the PDS signal was detected and the time when the voltage Vac is passed to zero (or other reference value). In such an embodiment, the notion of Time slot is implicit in determining a correlation between the PDS signals, since the amplitude of the voltage Vac is directly a function of the time after each zero crossing.

In yet another embodiment, the zero crossing detection circuit ZCCT is replaced by a measurement circuit VMCT1 configured to measure the voltage Vac and define groups VGi of values of the voltage Vac, for example 25 groups of values VG0. at VG24 with a range of 20V each, as shown below: vg0 VG1 ... VG11 VG12 VG13 VG14 ... VG23 VG24 0-20 20-40 ... 200-220 220-240 240-220 220-200 ... 40-20 20-0

Information corresponding to the groups of voltage values is supplied to the processing circuit MC in real time. The processing circuit searches for a correlation between the PDS signals by using these groups of values in place of the time slots.

The groups of voltage values of the voltage Vac could also have a variable voltage increment, for example 20V up to 100V, 10V between 100 and 200V and 5V between 200 and 240V. Similarly, the previously described time slots could have a variable time increment.

In yet another embodiment, the zero crossing detection circuit ZCCT is replaced by a measuring circuit VMCT2 configured to measure the voltage Vac and supply in real time to the processing circuit MC a parameter indicative of the value of the voltage. Vac, for example a binary signal coded on several bits. The processing circuit associates with each detected PDS signal the received parameter parameter at the moment when the PDS signal has been detected, this parameter then forming a parameter indicative of the position of the PDS signal on the curve of the voltage Vac. The processing circuit then seeks a correlation between the PDS signals using this indicative parameter instead of groups of voltage values of the voltage Vac or time slots.

Thus, in the present application, the notions of time slots, groups of voltage values of the AC voltage Vac or parameter indicative of the position of a PDS signal on the curve of the AC voltage Vac are considered equivalent. If the voltage Vac is sinusoidal, time slots of fixed duration correspond to groups of voltage values of the voltage Vac of variable extent, and groups of voltage values of the voltage Vac of fixed range correspond to time slots. variable duration.

The present invention is also susceptible of various applications. Thus, identical transient phenomena can be observed in a photovoltaic panel installation comprising diodes having an insulation fault. A detection method according to the invention also applies to this type of electrical network and can allow an improvement in the maintenance of energy production facilities. In this case, the correlation rules specific to this type of insulation fault are defined, always in relation to the position of the partial discharge pulses on the curve of the AC voltage conveyed by the monitored network.

Claims (15)

1. Device (DTC) for detecting, in an electrical network carrying a low-voltage AC voltage (Vac), an insulation fault between a phase wire (P) and a neutral wire (N) manifested in the appearance of partial discharge signals (PDS), comprising means for detecting partial discharge signals by analysing the voltage (Vac) or the current carried by the electrical network, characterized in that it comprises means for:

   - associating, with each detected partial discharge signal, a parameter (SLi, VGi) indicative of the position of the partial discharge signal on the curve of the AC voltage, and

   - searching for the existence, among a plurality of partial discharge signals detected during different variation cycles of the AC voltage, of at least one correlation (COR1, COR2, COR3) indicative of an insulation fault, taking into account the parameter (SLi, VGi) indicative of the position of the partial discharge signal on the curve of the AC voltage.
2. Device according to Claim 1, configured to:

   - define, as a parameter indicative of the position of a partial discharge signal on the curve of the AC voltage (Va), groups (SLi, VGi) of voltage values of the AC voltage, and

   - associate, with each detected discharge signal, a group (SLi, VGi) of voltage values of the AC voltage.
3. Device according to Claim 2, configured to identify each group (VGi) of voltage values of the AC voltage by means of a time slot (SLi) representing a fraction of the period (T) of the AC voltage.
4. Device according to either of Claims 2 and 3, configured to detect an insulation fault when a first number (N1) of partial discharge signals is detected during a first number (P1) of successive cycles of the AC voltage in relation to the same group (SLi, VGi) of voltage values of the AC voltage.
5. Device according to one of Claims 2 to 4, configured to detect an insulation fault when:

   - a first number (N1) of partial discharge signals is detected during a first number (P1) of successive cycles of the AC voltage in relation to a first group of voltage values of the AC voltage, and

   - a second number (N2) of partial discharge signals is then detected during a second number (P2) of successive cycles of the AC voltage in relation to a second group of voltage values of the AC voltage comprising voltage values lower than the values of the first group.
6. Device according to one of Claims 1 to 5, wherein the means for detecting partial discharge signals comprise:

   - an acquisition circuit (ACT) for extracting, from the AC voltage, a reference signal (Ve3) possibly containing partial discharge signals,

   - a circuit (MC) for processing the reference signal, configured, for each group (SLi, VGi) of voltage values of the AC voltage (Vac), to:

   - determine a current value (Vi) of the amplitude of the reference signal (Ve3),

   - calculate a mean value (MVi) of the amplitude of the envelope reference signal on the basis of the current value (Vi) of the amplitude of the reference signal and M-1 previous values of the amplitude of the reference signal that were measured over previous cycles of the AC voltage in relation to the group of voltage values of the AC voltage in question,

   - determine a detection threshold (Si, S1-S9) that depends on the mean value (MVi) of the amplitude of the reference signal, and

   - detect a partial discharge signal if the current value (Vi) of the amplitude of the reference signal is higher than the detection threshold (Si, S1-S9).
7. Device according to Claim 6, wherein the acquisition circuit (ACT) comprises an integrator circuit (EECT, D1, R1, C3) receiving as input a filtered signal (Ve2) extracted from the AC voltage and delivering the reference signal (Ve3).
8. Device according to either of Claims 6 and 7, wherein the detection circuit (MC) is configured to digitize the reference signal (Ve3) for each group (SLi, VGi) of voltage values of the AC voltage, and to deliver a digital sample (Vi) representative of the amplitude of the reference signal (Ve3) in relation to the group of voltage values of the AC voltage in question.
9. Device according to Claim 8, wherein the detection circuit (MC) is configured to calculate the mean value (MVi) of the amplitude of the reference signal (Ve3) on the basis of a group of M digital samples (Vi) comprising the digital sample (Vi) representative of the current amplitude of the reference signal (Ve3) and M-1 digital samples (Vi) representative of the amplitude of the reference signal (Ve3) over previous cycles of the AC voltage in relation to the same group of voltage values of the AC voltage.
10. Device according to one of Claims 6 to 9, wherein the detection circuit is configured to generate binary words (Wj) of N bits (b0-b9) in which each bit is associated with a group (SLi, VGi) of voltage values of the AC voltage, and has a first value if a partial discharge signal has been detected in relation to the group of voltage values of the AC voltage in question and a second logic value if a partial discharge signal has not been detected in relation to the group of voltage values in question.
11. Method for detecting an insulation fault between a phase wire (P) and a neutral wire (N) in an electrical network carrying a low-voltage AC voltage (Vac), comprising steps of detecting partial discharge signals by analysing the voltage (Vac) or the current carried by the electrical network, characterized in that it comprises steps consisting in:

    - associating, with each detected partial discharge signal, a parameter (SLi, VGi) indicative of the position of the partial discharge signal on the curve of the AC voltage, and

    - searching for the existence, among a plurality of partial discharge signals detected during different variation cycles of the AC voltage, of at least one correlation (COR1, COR2, COR3) indicative of an insulation fault, taking into account the parameter (SLi, VGi) indicative of the position of the partial discharge signal on the curve of the AC voltage.
12. Method according to Claim 11, comprising steps consisting in:

    - defining, as a parameter indicative of the position of a partial discharge signal on the curve of the AC voltage, groups (VGi) of voltage values of the AC voltage, and

    - associating, with each detected discharge signal, a group (VGi) of voltage values of the AC voltage.
13. Method according to Claim 12, wherein each group (VGi) of voltage values of the AC voltage is identified by means of a time slot (SLi) representing a fraction of the period (T) of the AC voltage.
14. Method according to either of Claims 12 and 13, wherein an insulation fault is detected when a first number (N1) of partial discharge signals is detected during a first number (P1) of successive cycles of the AC voltage in relation to the same group (SLi, VGi) of voltage values of the AC voltage.
15. Method according to one of Claims 12 to 14, wherein an insulation fault is detected when:

    - a first number (N1) of partial discharge signals is detected during a first number (P1) of successive cycles of the AC voltage in relation to a first group of voltage values of the AC voltage, and

    - a second number (N2) of partial discharge signals is then detected during a second number (P2) of successive cycles of the AC voltage in relation to a second group of voltage values of the AC voltage comprising voltage values lower than the values of the first group.

Images